Summary:
A number of optimizations are inhibited by single-use TokenFactors not
being merged into the TokenFactor using it. This makes we consider if
we can do the merge immediately.
Most tests changes here are due to the change in visitation causing
minor reorderings and associated reassociation of paired memory
operations.
CodeGen tests with non-reordering changes:
X86/aligned-variadic.ll -- memory-based add folded into stored leaq
value.
X86/constant-combiners.ll -- Optimizes out overlap between stores.
X86/pr40631_deadstore_elision -- folds constant byte store into
preceding quad word constant store.
Reviewers: RKSimon, craig.topper, spatel, efriedma, courbet
Reviewed By: courbet
Subscribers: dylanmckay, sdardis, nemanjai, jvesely, nhaehnle, javed.absar, eraman, hiraditya, kbarton, jrtc27, atanasyan, jsji, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D59260
llvm-svn: 356068
First step towards PR40800 - I intend to move the float case in a separate future patch.
I had to tweak the (overly reduced) thumb2 test and the x86 widening test change is annoying (no longer rematerializable) but we should address this separately.
Differential Revision: https://reviews.llvm.org/D59244
llvm-svn: 356040
The existing statepoint lowering code does something odd; it adds machine memory operands post instruction selection. This was copied from the stackmap/patchpoint implementation, but appears to be non-idiomatic.
This change is largely NFC. It moves the MMO creation logic into SelectionDAG building. It ends up not quite being NFC because the size of the stack slot is reflected in the MMO. The old code blindly used pointer size for the MMO size, which appears to have always been incorrect for larger values. It just happened nothing actually relied on the MMOs, so it worked out okay.
For context, I'm planning on removing the MOVolatile flag from these in a future commit, and then removing the MOStore flag from deopt spill slots in a separate one. Doing so is motivated by a small test case where we should be able to better schedule spill slots, but don't do so due to a memory use/def implied by the statepoint.
Differential Revision: https://reviews.llvm.org/D59106
llvm-svn: 355953
Expand MULO with constant power of two operand into a shift. The
overflow is checked with (x << shift) >> shift == x, where the right
shift will be logical for umulo and arithmetic for smulo (with
exception for multiplications by signed_min).
Differential Revision: https://reviews.llvm.org/D59041
llvm-svn: 355937
Fixes https://bugs.llvm.org/show_bug.cgi?id=36796.
Implement basic legalizations (PromoteIntRes, PromoteIntOp,
ExpandIntRes, ScalarizeVecOp, WidenVecOp) for VECREDUCE opcodes.
There are more legalizations missing (esp float legalizations),
but there's no way to test them right now, so I'm not adding them.
This also includes a few more changes to make this work somewhat
reasonably:
* Add support for expanding VECREDUCE in SDAG. Usually
experimental.vector.reduce is expanded prior to codegen, but if the
target does have native vector reduce, it may of course still be
necessary to expand due to legalization issues. This uses a shuffle
reduction if possible, followed by a naive scalar reduction.
* Allow the result type of integer VECREDUCE to be larger than the
vector element type. For example we need to be able to reduce a v8i8
into an (nominally) i32 result type on AArch64.
* Use the vector operand type rather than the scalar result type to
determine the action, so we can control exactly which vector types are
supported. Also change the legalize vector op code to handle
operations that only have vector operands, but no vector results, as
is the case for VECREDUCE.
* Default VECREDUCE to Expand. On AArch64 (only target using VECREDUCE),
explicitly specify for which vector types the reductions are supported.
This does not handle anything related to VECREDUCE_STRICT_*.
Differential Revision: https://reviews.llvm.org/D58015
llvm-svn: 355860
Includes a fix to emit a CheckOpcode for build_vector when immAllZerosV/immAllOnesV is used as a pattern root. This means it can't be used to look through bitcasts when used as a root, but that's probably ok. This extra CheckOpcode will ensure that the first match in the isel table will be a SwitchOpcode which is needed by the caching optimization in the ISel Matcher.
Original commit message:
Previously we had build_vector PatFrags that called ISD::isBuildVectorAllZeros/Ones. Internally the ISD::isBuildVectorAllZeros/Ones look through bitcasts, but we aren't able to take advantage of that in isel. Instead of we have to canonicalize the types of the all zeros/ones build_vectors and insert bitcasts. Then we have to pattern match those exact bitcasts.
By emitting specific matchers for these 2 nodes, we can make isel look through any bitcasts without needing to explicitly match them. We should also be able to remove the canonicalization to vXi32 from lowering, but I've left that for a follow up.
This removes something like 40,000 bytes from the X86 isel table.
Differential Revision: https://reviews.llvm.org/D58595
llvm-svn: 355784
This avoids breaking possible value dependencies when sorting loads by
offset.
AMDGPU has some load instructions that write into the high or low bits
of the destination register, and have a tied input for the other input
bits. These can easily have the same base pointer, but be a swizzle so
the high address load needs to come first. This was inserting glue
forcing the opposite ordering, producing a cycle the InstrEmitter
would assert on. It may be potentially expensive to look for the
dependency between the other loads, so just skip any where this could
happen.
Fixes bug 40936 by reverting r351379, which added a hacky attempt to
fix this by adding chains in this case, which I think was just working
around broken glue before the InstrEmitter. The core of the patch is
re-implementing the fix for that problem.
llvm-svn: 355728
Move the x86 combine from D58974 into the DAGCombine VSELECT code and update the SELECT version to use the isBooleanFlip helper as well.
Requested by @spatel on D59006
llvm-svn: 355533
During the lowering of a switch that would result in the generation of a
jump table, a range check is performed before indexing into the jump
table, for the switch value being outside the jump table range and a
conditional branch is inserted to jump to the default block. In case the
default block is unreachable, this conditional jump can be omitted. This
patch implements omitting this conditional branch for unreachable
defaults.
Differential Revision: https://reviews.llvm.org/D52002
Reviewers: Hans Wennborg, Eli Freidman, Roman Lebedev
llvm-svn: 355490
During the lowering of a switch that would result in the generation of a
jump table, a range check is performed before indexing into the jump
table, for the switch value being outside the jump table range and a
conditional branch is inserted to jump to the default block. In case the
default block is unreachable, this conditional jump can be omitted. This
patch implements omitting this conditional branch for unreachable
defaults.
Differential Revision: https://reviews.llvm.org/D52002
Reviewers: Hans Wennborg, Eli Freidman, Roman Lebedev
llvm-svn: 355483
This caused the first matcher in the isel table for many targets to Opc_Scope instead of Opc_SwitchOpcode. This leads to a significant increase in isel match failures.
llvm-svn: 355433
This patch enables combining integer bitcasts of integer build vectors when the new scalar type is legal. I've avoided floating point because the implementation bitcasts float to int along the way and we would need to check the intermediate types for legality
Differential Revision: https://reviews.llvm.org/D58884
llvm-svn: 355324
Summary:
Before when we implemented the first EH proposal, 'catch <tag>'
instruction may not catch an exception so there were multiple EH pads an
exception can unwind to. That means a BB could have multiple EH pad
successors.
Now after we switched to the new proposal, every 'catch' instruction
catches an exception, and there is only one catchpad per catchswitch, so
we at most have one EH pad successor, making `ThrowUnwindDest` map in
`WasmEHInfo` unnecessary.
Keeping `ThrowUnwindDest` map in `WasmEHInfo` has its own problems,
because other optimization passes can split a BB that contains possibly
throwing calls (previously invokes), and we have to update the map every
time that happens, which is not easy for common CodeGen passes.
This also correctly updates successor info in LateEHPrepare when we add
a rethrow instruction.
Reviewers: dschuff
Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D58486
llvm-svn: 355296
Previously we had build_vector PatFrags that called ISD::isBuildVectorAllZeros/Ones. Internally the ISD::isBuildVectorAllZeros/Ones look through bitcasts, but we aren't able to take advantage of that in isel. Instead of we have to canonicalize the types of the all zeros/ones build_vectors and insert bitcasts. Then we have to pattern match those exact bitcasts.
By emitting specific matchers for these 2 nodes, we can make isel look through any bitcasts without needing to explicitly match them. We should also be able to remove the canonicalization to vXi32 from lowering, but I've left that for a follow up.
This removes something like 40,000 bytes from the X86 isel table.
Differential Revision: https://reviews.llvm.org/D58595
llvm-svn: 355224
Summary:
The description of KnownBits::zext() and
KnownBits::zextOrTrunc() has confusingly been telling
that the operation is equivalent to zero extending the
value we're tracking. That has not been true, instead
the user has been forced to explicitly set the extended
bits as known zero afterwards.
This patch adds a second argument to KnownBits::zext()
and KnownBits::zextOrTrunc() to control if the extended
bits should be considered as known zero or as unknown.
Reviewers: craig.topper, RKSimon
Reviewed By: RKSimon
Subscribers: javed.absar, hiraditya, jdoerfert, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D58650
llvm-svn: 355099
At the moment, we mark every atomic memory access as being also volatile. This is unnecessarily conservative and prohibits many legal transforms (DCE, folding, etc..).
This patch removes MOVolatile from the MachineMemOperands of atomic, but not volatile, instructions. This should be strictly NFC after a series of previous patches which have gone in to ensure backend code is conservative about handling of isAtomic MMOs. Once it's in and baked for a bit, we'll start working through removing unnecessary bailouts one by one. We applied this same strategy to the middle end a few years ago, with good success.
To make sure this patch itself is NFC, it is build on top of a series of other patches which adjust code to (for the moment) be as conservative for an atomic access as for a volatile access and build up a test corpus (mostly in test/CodeGen/X86/atomics-unordered.ll)..
Previously landed
D57593 Fix a bug in the definition of isUnordered on MachineMemOperand
D57596 [CodeGen] Be conservative about atomic accesses as for volatile
D57802 Be conservative about unordered accesses for the moment
rL353959: [Tests] First batch of cornercase tests for unordered atomics.
rL353966: [Tests] RMW folding tests w/unordered atomic operations.
rL353972: [Tests] More unordered atomic lowering tests.
rL353989: [SelectionDAG] Inline a single use helper function, and remove last non-MMO interface
rL354740: [Hexagon, SystemZ] Be super conservative about atomics
rL354800: [Lanai] Be super conservative about atomics
rL354845: [ARM] Be super conservative about atomics
Attention Out of Tree Backend Owners: This patch may break you. If it does, you can use the TLI getMMOFlags hook to restore the MOVolatile to any instruction you need to. (See llvm-dev thread titled "PSA: Changes to how atomics are handled in backends" started Feb 27, 2019.)
Differential Revision: https://reviews.llvm.org/D57601
llvm-svn: 355025
If SADDSAT/SSUBSAT are legal, then we can expand SADDO/SSUBO by performing a ADD/SUB and a SADDO/SSUBO and then compare the results.
I looked at doing this for UADDO/USUBO as well but as we don't have to do as many range comparisons I didn't see any/much benefit.
Differential Revision: https://reviews.llvm.org/D58637
llvm-svn: 354866
These helpers extend the existing isConstOrConstSplat helper checks to support DemandedElts masks as well.
We already had a local version of this in SelectionDAG that computeKnownBits/ComputeNumSignBits made use of, but this adds the functionality directly to the BuildVectorSDNode node and extends isConstOrConstSplat etc. to use that.
This will allow us to reuse the functionality in SimplifyDemandedVectorElts/SimplifyDemandedBits.
Differential Revision: https://reviews.llvm.org/D58503
llvm-svn: 354797
Support undef shuffle mask indices in the shuffle(concat_vectors, concat_vectors) -> concat_vectors fold
Differential Revision: https://reviews.llvm.org/D58585
llvm-svn: 354793
OPC_CheckCondCode is always used as operand 2 of a setcc. And its always surrounded by a MoveChild2 and a MoveParent. By having a dedicated opcode for this case we can reduce the number of bytes needed for this pattern from 4 bytes to 2.
This saves ~3000 bytes in the X86 table.
llvm-svn: 354763
Summary:
When promoting the over flow vector for these ops we should use the target's desired setcc result type. This way a v8i32 result type will use a v8i32 overflow vector instead of a v8i16 overflow vector. A v8i16 overflow vector will cause LegalizeDAG/LegalizeVectorOps to have to use v8i32 and truncate to v8i16 in its expansion. By doing this in type legalization instead, we get the truncate into the DAG earlier and give DAG combine more of a chance to optimize it.
We also have to fix unrolling to use the scalar setcc result type for the scalarized operation, and convert it to the required vector element type after the scalar operation. We have to observe the vector boolean contents when doing this conversion. The previous code was just taking the scalar result and putting it in the vector. But for X86 and AArch64 that would have only put a the boolean value in bit 0 of the element and left all other bits in the element 0. We need to ensure all bits in the element are the same. I'm using a select with constants here because that's what setcc unrolling in LegalizeVectorOps used.
Reviewers: spatel, RKSimon, nikic
Reviewed By: nikic
Subscribers: javed.absar, kristof.beyls, dmgreen, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D58567
llvm-svn: 354753
r354648 was a follow up to fix a regression "[X86] Add a DAG combine for (aext_vector_inreg (aext_vector_inreg X)) -> (aext_vector_inreg X) to fix a regression from my previous commit."
These were reverted in r354713 as their context depended on other patches that were reverted for a bug.
llvm-svn: 354734
r354363 caused https://crbug.com/934963#c1, which has a plain C reduced
test case.
I also had to revert some dependent changes:
- r354648
- r354647
- r354640
- r354511
llvm-svn: 354713
When we need to merge two adjacent loads the AND mask for the low piece was still sized for the full src element size. But we didn't have that many bits. The upper bits are already zero due to the SRL. So we can skip the AND if we're going to combine with the high bits.
We do need an AND to clear out any bits from the high part. We were anding the high part before combining with the low part, but it looks like ANDing after the OR gets better results.
So we can just emit the final AND after the optional concatentation is done. That will handling skipping before the OR and get rid of extra high bits after the OR.
llvm-svn: 354655
Otherwise we end up creating extract_vector_elts that then each need to have their input promoted. This can lead to truncates needing to be emitted for each of those.
But we already emitted any_extends when we legalized the extract_subvector. So now we have pairs of any_extend+trunc that partially cancel. But depending on how DAGCombiner visits them we can get weird results.
By promoting the input at the same time we can create only a single any_extend or truncate.
There's one regression in the vector-narrow-binop.ll case, but that looks easy to fix with a follow up patch.
llvm-svn: 354647
This fold can occur during legalization, so it can fight with promotion
to the larger type. It apparently takes a special sequence and subtarget
to avoid more basic simplifications that would hide the problem.
But there's a bigger question raised here: why does distributeTruncateThroughAnd()
even exist? It duplicates functionality from a more minimal pattern that we
already have. But getting rid of this function requires some preliminary steps.
https://bugs.llvm.org/show_bug.cgi?id=40793
llvm-svn: 354594
If the LHS has known zeros, then the RHS immediate mask might have been simplified to remove those bits.
This patch adds a call to computeKnownBits to get the known zeroes to handle that possibility. I left an early out to skip the call if all of the demanded bits are set in the mask.
Differential Revision: https://reviews.llvm.org/D58464
llvm-svn: 354514
Second part of https://bugs.llvm.org/show_bug.cgi?id=40442.
This adds an extra UnrollVectorOverflowOp() method to SDAG, because
the general UnrollOverflowOp() method can't deal with multiple results.
Additionally we need to expand UMULO/SMULO during vector op
legalization, as it may result in unrolling, which may need additional
type legalization.
Differential Revision: https://reviews.llvm.org/D57997
llvm-svn: 354513
If the bit position has known zeros in it, then the AND immediate will likely be optimized to remove bits.
This can prevent GetDemandedBits from recognizing that the AND is unnecessary.
llvm-svn: 354498
Directly use the correct shift amount type if it is possible, and
future-proof the code against vectors. The added test makes sure that
bitwidths that do not fit into the shift amount type do not assert.
Split out from D57997.
llvm-svn: 354359
Summary:
A store to an object whose lifetime is about to end can be removed.
See PR40550 for motivation.
Reviewers: niravd
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D57541
llvm-svn: 354244
In preparation for supporting vector expansion.
Add an isPostTypeLegalization flag to makeLibCall(), because this
expansion relies on the legalized form using MERGE_VALUES. Drop
the corresponding variant of ExpandLibCall, which is no longer used.
Differential Revision: https://reviews.llvm.org/D58006
llvm-svn: 354226
While rebasing a refactor in r353950 I accidentally swapped two function
arguments; one is SelectionDAGBuilders "current" DebugLoc, the other is the one
from the "current" debug intrinsic. They're probably always identical, but I
haven't proved that yet.
llvm-svn: 354019
For D57601, we need to know whether the instruction is volatile. We'd either have to pass yet another parameter, or just standardize on the MMO interface. I chose the second.
llvm-svn: 353989
The helper function was used by only two callers, and largely ended up providing distinct functionality based on optional arguments and opcode. Inline and simply to make the functionality much more clear.
llvm-svn: 353977
In this patch SelectionDAG tries to salvage any dbg.values that are going to be
dropped, in case they can be recovered from Values in the current BB. It also
strengthens SelectionDAGs handling of dangling debug data, so that dbg.values
are *always* emitted (as Undef or otherwise) instead of dangling forever.
The motivation behind this patch exists in the new test case: a memory address
(here a bitcast and GEP) exist in one basic block, and a dbg.value referring to
the address is left in the 'next' block. The base pointer is live across all
basic blocks. In current llvm trunk the dbg.value cannot be encoded, and it
isn't even emitted as an Undef DBG_VALUE.
The change is simply: if we're definitely going to drop a dbg.value, repeatedly
apply salvageDebugInfo to its operand until either we find something that can
be encoded, or we can't salvage any further in which case we produce an Undef
DBG_VALUE. To know when we're "definitely going to drop a dbg.value",
SelectionDAG signals SelectionDAGBuilder when all IR instructions have been
encoded to force salvaging. This ensures that any dbg.value that's dangling
after DAG creation will have a corresponding DBG_VALUE encoded.
Differential Revision: https://reviews.llvm.org/D57694
llvm-svn: 353954
This is a pure copy-and-paste job, moving the logic for lowering dbg.value
intrinsics to SDDbgValues into its own function. This is ahead of adding some
more users of this logic.
Differential Revision: https://reviews.llvm.org/D57697
llvm-svn: 353950
SelectionDAGBuilder has special handling for dbg.value intrinsics that are
understood to define the location of function parameters on entry to the
function. To enable this, we avoid recording a dbg.value as a virtual register
reference if it might be such a parameter, so that it later hits
EmitFuncArgumentDbgValue.
This patch reduces the set of circumstances where we avoid recording a
dbg.value as a virtual register reference, to allow more "normal" variables
to be recorded that way. We now only bypass for potential parameters if:
* The dbg.value operand is an Argument,
* The Variable is a parameter, and
* The Variable is not inlined.
meaning it's very likely that the dbg.value is a function-entry parameter
location.
Differential Revision: https://reviews.llvm.org/D57584
llvm-svn: 353948
If we're comparing some value for equality against 2 constants
and those constants have an absolute difference of just 1 bit,
then we can offset and mask off that 1 bit and reduce to a single
compare against zero:
and/or (setcc X, C0, ne), (setcc X, C1, ne/eq) -->
setcc ((add X, -C1), ~(C0 - C1)), 0, ne/eq
https://rise4fun.com/Alive/XslKj
This transform is disabled by default using a TLI hook
("convertSetCCLogicToBitwiseLogic()").
That should be overridden for AArch64, MIPS, Sparc and possibly
others based on the asm shown in:
https://bugs.llvm.org/show_bug.cgi?id=40611
llvm-svn: 353859
Summary:
The SMULO/UMULO DAG nodes, when not directly supported by the target,
expand to a multiplication twice as wide. In case that the resulting
type is not legal, the legalizer cannot directly call the intrinsic
with the wide arguments; instead, it "pre-lowers" them by splitting
them in halves.
rL283203 made sure that on big endian targets, the legalizer passes
the argument halves in the correct order. It did not do the same
for the return value halves because the existing code used a hack;
it put an illegal type into DAG and hoped that nothing would break
and it would be correctly lowered elsewhere.
rL307207 fixed this, handling return value halves similar to how
argument handles are handled, but did not take big-endian targets
into account.
This commit fixes the expansion on big-endian targets, such as
the out-of-tree OR1K target.
Reviewers: eli.friedman, vadimcn
Subscribers: george-hopkins, efriedma, llvm-commits
Differential Revision: https://reviews.llvm.org/D45355
llvm-svn: 353854
Summary:
This patch fixes PR40587.
When a dbg.value instrinsic is emitted to the DAG
by using EmitFuncArgumentDbgValue the resulting
DBG_VALUE is hoisted to the beginning of the entry
block. I think the idea is to be able to locate
a formal argument already from the start of the
function.
However, EmitFuncArgumentDbgValue only checked that
the value that was used to describe a variable was
originating from a function parameter, not that the
variable itself actually was an argument to the
function. So when for example assigning a local
variable "local" the value from an argument "a",
the assocated DBG_VALUE instruction would be hoisted
to the beginning of the function, even if the scope
for "local" started somewhere else (or if "local"
was mapped to other values earlier in the function).
This patch adds some logic to EmitFuncArgumentDbgValue
to check that the variable being described actually
is an argument to the function. And that the dbg.value
being lowered already is in the entry block. Otherwise
we bail out, and the dbg.value will be handled as an
ordinary dbg.value (not as a "FuncArgumentDbgValue").
A tricky situation is when both the variable and
the value is related to function arguments, but not
neccessarily the same argument. We make sure that we
do not describe the same argument more than once as
a "FuncArgumentDbgValue". This solution works as long
as opt has injected a "first" dbg.value that corresponds
to the formal argument at the function entry.
Reviewers: jmorse, aprantl
Subscribers: jyknight, hiraditya, fedor.sergeev, dstenb, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D57702
llvm-svn: 353735
`CallBase` class rather than `CallSite` wrappers.
I pushed this change down through most of the statepoint infrastructure,
completely removing the use of CallSite where I could reasonably do so.
I ended up making a couple of cut-points: generic call handling
(instcombine, TLI, SDAG). As soon as it hit truly generic handling with
users outside the immediate code, I simply transitioned into or out of
a `CallSite` to make this a reasonable sized chunk.
Differential Revision: https://reviews.llvm.org/D56122
llvm-svn: 353660
Now that we have vector support for [US](ADD|SUB)O we no longer
need to scalarize when expanding [US](ADD|SUB)SAT.
This matches what the cost model already does.
Differential Revision: https://reviews.llvm.org/D57348
llvm-svn: 353651
Now that we have SimplifyDemandedBits support for funnel shifts (rL353539), we need to simplify funnel shifts back to bitshifts in cases where either argument has been folded to undef/zero.
Differential Revision: https://reviews.llvm.org/D58009
llvm-svn: 353645
SimplifySetCC still has much room for improvement, but this should
fix the remaining problem examples from:
https://bugs.llvm.org/show_bug.cgi?id=40657
The initial fix for this problem was rL353615.
llvm-svn: 353639
There's effectively no difference for the cases with variables.
We just trade a sub for an add on those. But the case with a
subtract from constant would require an extra move instruction
on x86, so this looks like a reasonable generic combine.
llvm-svn: 353619
In preparation for supporting vector expansion.
Also drop a variant of ExpandLibCall, of which the MULO expansions
were the only user.
llvm-svn: 353611
This patch accompanies the RFC posted here:
http://lists.llvm.org/pipermail/llvm-dev/2018-October/127239.html
This patch adds a new CallBr IR instruction to support asm-goto
inline assembly like gcc as used by the linux kernel. This
instruction is both a call instruction and a terminator
instruction with multiple successors. Only inline assembly
usage is supported today.
This also adds a new INLINEASM_BR opcode to SelectionDAG and
MachineIR to represent an INLINEASM block that is also
considered a terminator instruction.
There will likely be more bug fixes and optimizations to follow
this, but we felt it had reached a point where we would like to
switch to an incremental development model.
Patch by Craig Topper, Alexander Ivchenko, Mikhail Dvoretckii
Differential Revision: https://reviews.llvm.org/D53765
llvm-svn: 353563
The sqrt case is faster and we already do this for the case where
the exponent is 0.25. This adds the 0.75 case which is also not
sensitive to signed zeros.
Patch by Whitney Tsang (Whitney)
Differential revision: https://reviews.llvm.org/D57434
llvm-svn: 353557
Replace OR(SHL,SRL) pattern with ISD::FSHR (legalization expands this later if necessary) - this helps with the scale == 0 'undefined' drop-through case that was discussed on D55720.
llvm-svn: 353546
This is part of https://bugs.llvm.org/show_bug.cgi?id=40442.
Vector legalization is implemented for the add/sub overflow opcodes.
UMULO/SMULO are also handled as far as legalization is concerned, but
they don't support vector expansion yet (so no tests for them).
The vector result widening implementation is suboptimal, because it
could result in a legalization loop.
Differential Revision: https://reviews.llvm.org/D57639
llvm-svn: 353464
Move the (add (umax X, C), -C) --> (usubsat X, C) X86 combine into generic DAGCombiner
First of a number of saturated arithmetic folds that can be moved out of X86-specific code for PR40111.
Differential Revision: https://reviews.llvm.org/D57754
llvm-svn: 353457
I noticed that we are missing this canonicalization in IR:
rL352515
...and then realized that we don't get this right in SDAG either,
so this has to be fixed first regardless of what we choose to do in IR.
The existing fold was limited to scalars and using the wrong predicate
to guard the transform. We have a boolean contents TLI query that can
be used to decide which direction to fold.
This may eventually lead back to the problems/question in:
https://bugs.llvm.org/show_bug.cgi?id=40486
...but it makes no difference to that yet.
Differential Revision: https://reviews.llvm.org/D57401
llvm-svn: 353433
Summary:
If the index isn't constant, this transform inserts a multiply and an add on the index to calculating the base pointer for a scalar load. But we still create a memory operand with an offset of 0 and the size of the scalar access. But the access is really to an unknown offset within the original access size.
This can cause the machine scheduler to incorrectly calculate dependencies between this load and other accesses. In the case we saw, there was a 32 byte vector store that was split into two 16 byte stores, one with offset 0 and one with offset 16. The size of the memory operand for both was 16. The scheduler correctly detected the alias with the offset 0 store, but not the offset 16 store.
This patch discards the pointer info so we don't incorrectly detect aliasing. I wasn't sure if we could keep using the original offset and size without risking some other transform on the load changing the size.
I tried to reduce a test case, but there's still a lot of memory operations needed to get the scheduler to do the bad reordering. So it looked pretty fragile to maintain.
Reviewers: efriedma
Reviewed By: efriedma
Subscribers: arphaman, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D57616
llvm-svn: 353124
Add an intrinsic that takes 2 unsigned integers with the scale of them
provided as the third argument and performs fixed point multiplication on
them.
This is a part of implementing fixed point arithmetic in clang where some of
the more complex operations will be implemented as intrinsics.
Differential Revision: https://reviews.llvm.org/D55625
llvm-svn: 353059
Noticed while investigating PR40483, and fixes the basic test case from the bug - but not a more general case.
We're pretty weak at dealing with ADD/SUB combines compared to the SimplifyAssociativeOrCommutative/SimplifyUsingDistributiveLaws abilities that InstCombine can manage.
llvm-svn: 353044
We already have the getConstantOperandVal helper which returns a uint64_t, but along comes the fuzzer and inserts a i128 -1 constant or something and the whole thing asserts.......
I've updated a few obvious cases, and tried to make use of the const reference where possible, but there's more to do. A number of existing oss-fuzz tickets should be fixed if we start using APInt and perform value clamping where necessary.
llvm-svn: 352961
Summary: This fixes using the correct stack registers for SEH when stack realignment is needed or when variable size objects are present.
Reviewers: rnk, efriedma, ssijaric, TomTan
Reviewed By: rnk, efriedma
Subscribers: javed.absar, kristof.beyls, llvm-commits
Differential Revision: https://reviews.llvm.org/D57183
llvm-svn: 352923
This cleans up all CallInst creation in LLVM to explicitly pass a
function type rather than deriving it from the pointer's element-type.
Differential Revision: https://reviews.llvm.org/D57170
llvm-svn: 352909
The version of FoldConstantArithmetic() that takes arbitrary nodes
was confusingly naming those nodes as constants when they might
not be; also "Cst" reads like "Cast".
llvm-svn: 352884
This might be the start of tracking all vector element constants generally if we take it to its
logical conclusion, but let's stop here and make sure this is correct/beneficial so far.
The affected tests require a convoluted path before they get simplified currently because we
don't call SimplifyDemandedVectorElts() from binops directly and don't modify the binop operands
directly in SimplifyDemandedVectorElts().
That's why the tests all have a trailing shuffle to induce a chain reaction of transforms. So
something like this is happening:
1. Improve the knowledge of undefs in the binop via a SimplifyDemandedVectorElts() call that
originates from a shuffle.
2. Transfer that undef knowledge back to the shuffle mask user as more undef lanes.
3. Combine the modified shuffle by calling SimplifyDemandedVectorElts() again.
4. Translate the improved shuffle mask as undemanded lanes of build vector constants causing
those to become full undef constants.
5. Simplify the binop now that it has a full undef operand.
As we can see from the unchanged 'and' and 'or' tests, tracking undefs alone isn't a full solution.
We would need to track zero and all-ones constants to improve those opcodes. We'd probably need to
track NaN for FP ops too (assuming we don't have fast-math-flags set).
Differential Revision: https://reviews.llvm.org/D57066
llvm-svn: 352880
For targets where i32 is not a legal type (e.g. 64-bit RISC-V),
LegalizeIntegerTypes must promote the integer operand of ISD::FPOWI. As this
is a signed value, this should be sign-extended.
This patch enables all tests in test/CodeGen/RISCVfloat-intrinsics.ll for
RV64, as prior to this patch that file couldn't be compiled for RV64 due to an
assertion when performing codegen for fpowi.
Differential Revision: https://reviews.llvm.org/D54574
llvm-svn: 352832
This patch fixes pr39098.
For the attached test case, CombineZExtLogicopShiftLoad can optimize it to
t25: i64 = Constant<1099511627775>
t35: i64 = Constant<0>
t0: ch = EntryToken
t57: i64,ch = load<(load 4 from `i40* undef`, align 8), zext from i32> t0, undef:i64, undef:i64
t58: i64 = srl t57, Constant:i8<1>
t60: i64 = and t58, Constant:i64<524287>
t29: ch = store<(store 5 into `i40* undef`, align 8), trunc to i40> t57:1, t60, undef:i64, undef:i64
But later visitANDLike transforms it to
t25: i64 = Constant<1099511627775>
t35: i64 = Constant<0>
t0: ch = EntryToken
t57: i64,ch = load<(load 4 from `i40* undef`, align 8), zext from i32> t0, undef:i64, undef:i64
t61: i32 = truncate t57
t63: i32 = srl t61, Constant:i8<1>
t64: i32 = and t63, Constant:i32<524287>
t65: i64 = zero_extend t64
t58: i64 = srl t57, Constant:i8<1>
t60: i64 = and t58, Constant:i64<524287>
t29: ch = store<(store 5 into `i40* undef`, align 8), trunc to i40> t57:1, t60, undef:i64, undef:i64
And it triggers CombineZExtLogicopShiftLoad again, causes a dead loop.
Both forms should generate same instructions, CombineZExtLogicopShiftLoad generated IR looks cleaner. But it looks more difficult to prevent visitANDLike to do the transform, so I prevent CombineZExtLogicopShiftLoad to do the transform if the ZExt is free.
Differential Revision: https://reviews.llvm.org/D57491
llvm-svn: 352792
While dangling nodes will eventually be pruned when they are
considered, leaving them disables combines requiring single-use.
Reviewers: Carrot, spatel, craig.topper, RKSimon, efriedma
Subscribers: hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D57520
llvm-svn: 352784
r zero scale SMULFIX, expand into MUL which produces better code for X86.
For vector arguments, expand into MUL if SMULFIX is provided with a zero scale.
Otherwise, expand into MULH[US] or [US]MUL_LOHI.
Differential Revision: https://reviews.llvm.org/D56987
llvm-svn: 352783
And instead just generate a libcall. My motivating example on ARM was a simple:
shl i64 %A, %B
for which the code bloat is quite significant. For other targets that also
accept __int128/i128 such as AArch64 and X86, it is also beneficial for these
cases to generate a libcall when optimising for minsize. On these 64-bit targets,
the 64-bits shifts are of course unaffected because the SHIFT/SHIFT_PARTS
lowering operation action is not set to custom/expand.
Differential Revision: https://reviews.llvm.org/D57386
llvm-svn: 352736
Summary:
Fixes PR40267, in which the removed assertion was triggering on
perfectly valid IR. As far as I can tell, constant out of bounds
indices should be allowed when splitting extract_vector_elt, since
they will simply be propagated as out of bounds indices in the
resulting split vector and handled appropriately elsewhere.
Reviewers: aheejin
Subscribers: dschuff, sbc100, jgravelle-google, hiraditya
Differential Revision: https://reviews.llvm.org/D57471
llvm-svn: 352702
These can be triggered by mistakenly using a 64-bit mode only intrinsics with a -mtriple=i686. Using report_fatal_error gives a better experience for this mistake in release builds instead of probably crashing.
We already do this for some of the vector type legalization handles.
llvm-svn: 352699
This extends the existing transform for:
add X, 0/1 --> sub X, 0/-1
...to allow the sibling subtraction fold.
This pattern could regress with the proposed change in D57401.
llvm-svn: 352680
Summary:
This switches the EH implementation to the new proposal:
https://github.com/WebAssembly/exception-handling/blob/master/proposals/Exceptions.md
(The previous proposal was
https://github.com/WebAssembly/exception-handling/blob/master/proposals/old/Exceptions.md)
- Instruction changes
- Now we have one single `catch` instruction that returns a except_ref
value
- `throw` now can take variable number of operations
- `rethrow` does not have 'depth' argument anymore
- `br_on_exn` queries an except_ref to see if it matches the tag and
branches to the given label if true.
- `extract_exception` is a pseudo instruction that simulates popping
values from wasm stack. This is to make `br_on_exn`, a very special
instruction, work: `br_on_exn` puts values onto the stack only if it
is taken, and the # of values can vay depending on the tag.
- Now there's only one `catch` per `try`, this patch removes all special
handling for terminate pad with a call to `__clang_call_terminate`.
Before it was the only case there are two catch clauses (a normal
`catch` and `catch_all` per `try`).
- Make `rethrow` act as a terminator like `throw`. This splits BB after
`rethrow` in WasmEHPrepare, and deletes an unnecessary `unreachable`
after `rethrow` in LateEHPrepare.
- Now we stop at all catchpads (because we add wasm `catch` instruction
that catches all exceptions), this creates new
`findWasmUnwindDestinations` function in SelectionDAGBuilder.
- Now we use `br_on_exn` instrution to figure out if an except_ref
matches the current tag or not, LateEHPrepare generates this sequence
for catch pads:
```
catch
block i32
br_on_exn $__cpp_exception
end_block
extract_exception
```
- Branch analysis for `br_on_exn` in WebAssemblyInstrInfo
- Other various misc. changes to switch to the new proposal.
Reviewers: dschuff
Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D57134
llvm-svn: 352598
This is the sibling fold for insert-of-insert that was added with D56604.
Now that we have x86 shuffle narrowing (D57156), this change shows improvements for
lots of AVX512 reduction code (not sure that we would ever expect extract-of-extract otherwise).
There's a small regression in some of the partial-permute tests (extracting followed by splat).
That is tracked by PR40500:
https://bugs.llvm.org/show_bug.cgi?id=40500
Differential Revision: https://reviews.llvm.org/D57336
llvm-svn: 352528
This fixes most references to the paths:
llvm.org/svn/
llvm.org/git/
llvm.org/viewvc/
github.com/llvm-mirror/
github.com/llvm-project/
reviews.llvm.org/diffusion/
to instead point to https://github.com/llvm/llvm-project.
This is *not* a trivial substitution, because additionally, all the
checkout instructions had to be migrated to instruct users on how to
use the monorepo layout, setting LLVM_ENABLE_PROJECTS instead of
checking out various projects into various subdirectories.
I've attempted to not change any scripts here, only documentation. The
scripts will have to be addressed separately.
Additionally, I've deleted one document which appeared to be outdated
and unneeded:
lldb/docs/building-with-debug-llvm.txt
Differential Revision: https://reviews.llvm.org/D57330
llvm-svn: 352514
During the lowering of a switch that would result in the generation of a
jump table, a range check is performed before indexing into the jump
table, for the switch value being outside the jump table range and a
conditional branch is inserted to jump to the default block. In case the
default block is unreachable, this conditional jump can be omitted. This
patch implements omitting this conditional branch for unreachable
defaults.
Review ID: D52002
Reviewers: Hans Wennborg, Eli Freidman, Roman Lebedev
llvm-svn: 352484
A FrameIndex should be valid throughout a block regardless of what instructions
get selected in that block -- therefore we shouldn't harness dbg.values that
refer to FrameIndexes to an SDNode. There are numerous codegen reasons why
an SDNode never appears or doesn't become a location that a DBG_VALUE can
refer to. None of them actually affect the variable location.
Therefore, before any other tests to encode dbg_values in a SelectionDAG,
identify FrameIndex operands and encode them unattached to any SDNode.
Differential Revision: https://reviews.llvm.org/D57328
llvm-svn: 352467
This did not cause the buildbot failure it was previously reverted for.
Original commit message:
I'm not sure why we were using SEXTLOAD. EXTLOAD seems more appropriate since we don't care about the upper bits.
This patch changes this and then modifies the X86 post legalization combine to emit a extending shuffle instead of a sign_extend_vector_inreg. Could maybe use an any_extend_vector_inre
On AVX512 targets I think we might be able to use a masked vpmovzx and not have to expand this at all.
llvm-svn: 352433
Followup to D56636, this time handling the UADDSAT case by expanding
uadd.sat(a, b) to umin(a, ~b) + b.
Differential Revision: https://reviews.llvm.org/D56869
llvm-svn: 352409
This patch improves the placement of DBG_VALUEs when by SelectionDAG, which
as documented in PR40427 can go very wrong. At the core of this is
ProcessSourceNode, which assumes the last instruction in a BB is the start
of the last processed IR instruction, which isn't always true.
Instead, use a helper function to call InstrEmitter::EmitNode, that records
before-and-after iterators and determines the first of any new instruction
created during emission. This is passed to ProcessSourceNode, which can
then make more elightened decisions about ordering for DBG_VALUE placement.
Differential revision: https://reviews.llvm.org/D57163
llvm-svn: 352350
Summary:
I'm not sure why we were using SEXTLOAD. EXTLOAD seems more appropriate since we don't care about the upper bits.
This patch changes this and then modifies the X86 post legalization combine to emit a extending shuffle instead of a sign_extend_vector_inreg. Could maybe use an any_extend_vector_inreg, but I just did what we already do in LowerLoad. I think we can actually get rid of this code entirely if we switch to -x86-experimental-vector-widening-legalization.
On AVX512 targets I think we might be able to use a masked vpmovzx and not have to expand this at all.
Reviewers: RKSimon, spatel
Reviewed By: RKSimon
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D57186
llvm-svn: 352255
It should be emitted when any floating-point operations (including
calls) are present in the object, not just when calls to printf/scanf
with floating point args are made.
The difference caused by this is very subtle: in static (/MT) builds,
on x86-32, in a program that uses floating point but doesn't print it,
the default x87 rounding mode may not be set properly upon
initialization.
This commit also removes the walk of the types pointed to by pointer
arguments in calls. (To assist in opaque pointer types migration --
eventually the pointee type won't be available.)
That latter implies that it will no longer consider a call like
`scanf("%f", &floatvar)` as sufficient to emit _fltused on its
own. And without _fltused, `scanf("%f")` will abort with error R6002. This
new behavior is unlikely to bite anyone in practice (you'd have to
read a float, and do nothing with it!), and also, is consistent with
MSVC.
Differential Revision: https://reviews.llvm.org/D56548
llvm-svn: 352076
The current check in CombineToPreIndexedLoadStore is too
conversative, preventing a pre-indexed store when the base pointer
is a predecessor of the value being stored. Instead, we should check
the pointer operand of the store.
Differential Revision: https://reviews.llvm.org/D56719
llvm-svn: 351933
Also add debug prints in the default case of the switches in these routines.
Most if not all of the type legalization handlers already do this so this makes promoting floats consistent
llvm-svn: 351890
vecbo (insertsubv undef, X, Z), (insertsubv undef, Y, Z) --> insertsubv VecC, (vecbo X, Y), Z
This is another step in generic vector narrowing. It's also a step towards more horizontal op
formation specifically for x86 (although we still failed to match those in the affected tests).
The scalarization cases are also not optimal (we should be scalarizing those), but it's still
an improvement to use a narrower vector op when we know part of the result must be constant
because both inputs are undef in some vector lanes.
I think a similar match but checking for a constant operand might help some of the cases in
D51553.
Differential Revision: https://reviews.llvm.org/D56875
llvm-svn: 351825
The regression test is reduced from the example shown in D56281.
This does raise a question as noted in the test file: do we want
to handle this pattern? I don't have a motivating example for
that on x86 yet, but it seems like we could have that pattern
there too, so we could avoid the back-and-forth using a shuffle.
llvm-svn: 351753
to reflect the new license.
We understand that people may be surprised that we're moving the header
entirely to discuss the new license. We checked this carefully with the
Foundation's lawyer and we believe this is the correct approach.
Essentially, all code in the project is now made available by the LLVM
project under our new license, so you will see that the license headers
include that license only. Some of our contributors have contributed
code under our old license, and accordingly, we have retained a copy of
our old license notice in the top-level files in each project and
repository.
llvm-svn: 351636
Summary:
This patch makes some changes related to -dag-dump-verbose.
Main use case has been when debugging how SelectionDAG is
dealing with debug info (SDDbgValue nodes).
1) We now print the number of DbgValues that are mapped to each
SDNode.
2) Removed duplicated printing of DebugLoc (nowadays DebugLoc is
printed also when not using -dag-dump-verbose).
3) Renamed SDDbgValue::dump to SDDbgValue::print, and added a
new SDDbgValue::dump that will start a new line after calling
print.
4) SDDbgValue::print now prints "Order", and it also prints
some additional information when kind is CONST/FRAMEIX/VREG.
5) SelectionDAG::dump() now dumps all SDDbgValue nodes after
the list of SDNodes (both "regular" and "ByVal" SDDbgValue:s).
Invalidated nodes are not printed.
6) Prohibit inline printing of SDNode operands that has SDDbgValue
nodes associated to them.
Reviewers: jmorse, aprantl
Reviewed By: aprantl
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D56793
llvm-svn: 351581
Similar to D55073. Without this change, the DAG combiner crashes on code
with more than 64k of stores in a single basic block that form parallelizable
chains.
No test case, as it would be very IR file.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D56740
llvm-svn: 351571
Defer inline asm's output fixup work until after we've generated the
inline asm node itself. Remove StoresToEmit, IndirectStoresToEmit, and
RetValRegs in favor of using ConstraintOperands.
llvm-svn: 351558
This functionality is required at multiple places which potentially
create large operand lists, like SelectionDAGBuilder or DAGCombiner.
Differential Revision: https://reviews.llvm.org/D56739
llvm-svn: 351552
Summary:
Use this helper to make sure we use the same value at various places.
This will likely be needed at more places were we currently crash
because we use more operands than possible.
Also makes it easier to change in the future.
Reviewers: RKSimon, craig.topper, efriedma, aemerson
Reviewed By: RKSimon
Subscribers: hiraditya, arsenm, llvm-commits
Differential Revision: https://reviews.llvm.org/D56859
llvm-svn: 351537
We should not pre-scheduled the node has ADJCALLSTACKDOWN parent,
or else, when bottom-up scheduling, ADJCALLSTACKDOWN and
ADJCALLSTACKUP may hold CallResource too long and make other
calls can't be scheduled. If there's no other available node
to schedule, the scheduler will try to rename the register by
creating copy to avoid the conflict which will fail because
CallResource is not a real physical register.
llvm-svn: 351527
Summary:
This patch supports MS SEH extensions __try/__except/__finally. The intrinsics localescape and localrecover are responsible for communicating escaped static allocas from the try block to the handler.
We need to preserve frame pointers for SEH. So we create a new function/property HasLocalEscape.
Reviewers: rnk, compnerd, mstorsjo, TomTan, efriedma, ssijaric
Reviewed By: rnk, efriedma
Subscribers: smeenai, jrmuizel, alex, majnemer, ssijaric, ehsan, dmajor, kristina, javed.absar, kristof.beyls, chrib, llvm-commits
Differential Revision: https://reviews.llvm.org/D53540
llvm-svn: 351370
dbg.value intrinsics can appear in blocks where their operand is not used,
meaning the operand never receives an SDNode, and thus no DBG_VALUE will
be created. Get around this by looking to see whether the operand has already
been allocated a virtual register. This allows dbg.values of Phi node and
Values that are used across basic blocks to successfully be translated into
DBG_VALUEs.
Differential Revision: https://reviews.llvm.org/D56678
llvm-svn: 351358
The value returned by max() is the last valid value, adjust the
comparison accordingly.
The code added in D55073 creates TokenFactors with max() operands.
Reviewers: aemerson, efriedma, RKSimon, craig.topper
Reviewed By: aemerson
Differential Revision: https://reviews.llvm.org/D56738
llvm-svn: 351318
ReduceLoadWidth can trigger using a shifted mask is used and this
requires that the function return a shl node to correct for the
offset. However, the way that this was implemented meant that the
returned result could be an existing node, which would be incorrect.
This fixes the method of inserting the new node and replacing uses.
Differential Revision: https://reviews.llvm.org/D50432
llvm-svn: 351310
The motivating case for this is shown in the first regression test. We are
transferring to scalar and back rather than just zero-extending with 'vpmovzxdq'.
That's a special-case for a more general pattern as shown here. In all tests,
we're avoiding the vector-scalar-vector moves in favor of vector ops.
We aren't producing optimal shuffle code in some cases though, so the patch is
limited to reduce regressions.
Differential Revision: https://reviews.llvm.org/D56281
llvm-svn: 351198
I accidentally triggered this code while doing some experiments and it doesn't look lke it could possibly work.
It calls 'getNOT' on a node that should be a CondCode.
I think to do this right we would need to swap the branch target and the fallthrough target. But that's not easy to do. Or we could create an explicit SetCC and feed that into a new BR_CC?
llvm-svn: 351022
This pattern:
t33: v8i32 = insert_subvector undef:v8i32, t35, Constant:i64<0>
t21: v16i32 = insert_subvector undef:v16i32, t33, Constant:i64<0>
...shows up in PR33758:
https://bugs.llvm.org/show_bug.cgi?id=33758
...although this patch doesn't make any difference to the final result on that yet.
In the affected tests here, it looks like it just makes RA wiggle. But we might
as well squash this to prevent it interfering with other pattern-matching.
Differential Revision:
https://reviews.llvm.org/D56604
llvm-svn: 351008
This patch takes some of the code from D49837 to allow us to enable ISD::ABS support for all SSE vector types.
Differential Revision: https://reviews.llvm.org/D56544
llvm-svn: 350998
Summary:
When legalizing the result of a SELECT_CC node by promoting the
floating-point type, use the promoted-to type rather than the original
type.
Fix PR40273.
Reviewers: efriedma, majnemer
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D56566
llvm-svn: 350951
That is, remove many of the calls to Type::getNumContainedTypes(),
Type::subtypes(), and Type::getContainedType(N).
I'm not intending to remove these accessors -- they are
useful/necessary in some cases. However, removing the pointee type
from pointers would potentially break some uses, and reducing the
number of calls makes it easier to audit.
llvm-svn: 350835
This removes check for single use from general ShrinkDemandedConstant
to the BE because of the AArch64 regression after D56289/rL350475.
After several hours of experiments I did not come up with a testcase
failing on any other targets if check is not performed.
Moreover, direct call to ShrinkDemandedConstant is not really needed
and superceed by SimplifyDemandedBits.
Differential Revision: https://reviews.llvm.org/D56406
llvm-svn: 350684
As we saw in D56057 when we tried to use this function on X86, it's unsafe. It allows the operand node to have multiple users, but doesn't prevent recursing past the first node when it does have multiple users. This can cause other simplifications earlier in the graph without regard to what bits are needed by the other users of the first node. Ideally all we should do to the first node if it has multiple uses is bypass it when its not needed by the user we started from. Doing any other transformation that SimplifyDemandedBits can do like turning ZEXT/SEXT into AEXT would result in an increase in instructions.
Fortunately, we already have a function that can do just that, GetDemandedBits. It will only make transformations that involve bypassing a node.
This patch changes AMDGPU's simplifyI24, to use a combination of GetDemandedBits to handle the multiple use simplifications. And then uses the regular SimplifyDemandedBits on each operand to handle simplifications allowed when the operand only has a single use. Unfortunately, GetDemandedBits simplifies constants more aggressively than SimplifyDemandedBits. This caused the -7 constant in the changed test to be simplified to remove the upper bits. I had to modify computeKnownBits to account for this by ignoring the upper 8 bits of the input.
Differential Revision: https://reviews.llvm.org/D56087
llvm-svn: 350560
The FSHL/FSHR nodes are handled in the expand function, but they need to also be listed in the code that queries for the operation action too.
llvm-svn: 350490
Fixes cvt_f32_ubyte combine. performCvtF32UByteNCombine() could shrink
source node to demanded bits only even if there are other uses.
Differential Revision: https://reviews.llvm.org/D56289
llvm-svn: 350475
This adds support for calculating sign bits of insert_subvector. I based it on the computeKnownBits.
My motivating case is propagating sign bits information across basic blocks on AVX targets where concatenating using insert_subvector is common.
Differential Revision: https://reviews.llvm.org/D56283
llvm-svn: 350432
As noted in PR39973 and D55558:
https://bugs.llvm.org/show_bug.cgi?id=39973
...this is a partial implementation of a fold that we do as an IR canonicalization in instcombine:
// extelt (binop X, Y), Index --> binop (extelt X, Index), (extelt Y, Index)
We want to have this in the DAG too because as we can see in some of the test diffs (reductions),
the pattern may not be visible in IR.
Given that this is already an IR canonicalization, any backend that would prefer a vector op over
a scalar op is expected to already have the reverse transform in DAG lowering (not sure if that's
a realistic expectation though). The transform is limited with a TLI hook because there's an
existing transform in CodeGenPrepare that tries to do the opposite transform.
Differential Revision: https://reviews.llvm.org/D55722
llvm-svn: 350354
Currently we expand the two nodes separately. This gives DAG combiner an opportunity to optimize the expanded sequence taking into account only one set of users. When we expand the other node we'll create the expansion again, but might not be able to optimize it the same way. So the nodes won't CSE and we'll have two similarish sequences in the same basic block. By expanding both nodes at the same time we'll avoid prematurely optimizing the expansion until both the division and remainder have been replaced.
Improves the test case from PR38217. There may be additional opportunities after this.
Differential Revision: https://reviews.llvm.org/D56145
llvm-svn: 350239
By also promoting the input type we get a better idea for what scalar type to use. This can provide better results if the result of the extract is sign extended. What was previously happening is that the extract result would be legalized, sometime later the input of the sign extend would be legalized using the result of the extract. Then later the extract input would be legalized forcing a truncate into the input of the sign extend using a replace all uses. This requires DAG combine to combine out the sext/truncate pair. But sometimes we visited the truncate first and messed things up before the sext could be combined.
By creating the extract with the correct scalar type when we create legalize the result type, the truncate will be added right away. Then when the sign_extend input is legalized it will create an any_extend of the truncate which can be optimized by getNode to maybe remove the truncate. And then a sign_extend_inreg. Now DAG combine doesn't have to worry about getting rid of the extend.
This fixes the regression on X86 in D56156.
Differential Revision: https://reviews.llvm.org/D56176
llvm-svn: 350236
If x has multiple sign bits than it doesn't matter which one we extend from so we can sext from x's msb instead.
The X86 setcc-combine.ll changes are a little weird. It appears we ended up with a (sext_inreg (aext (trunc (extractelt)))) after type legalization. The sext_inreg+aext now gets optimized by this combine to leave (sext (trunc (extractelt))). Then we visit the trunc before we visit the sext. This ends up changing the truncate to an extractvectorelt from a bitcasted vector. I have a follow up patch to fix this.
Differential Revision: https://reviews.llvm.org/D56156
llvm-svn: 350235
default
During the lowering of a switch that would result in the generation of a jump
table, a range check is performed before indexing into the jump table, for the
switch value being outside the jump table range and a conditional branch is
inserted to jump to the default block. In case the default block is
unreachable, this conditional jump can be omitted. This patch implements
omitting this conditional branch for unreachable defaults.
Review Reference: D52002
llvm-svn: 350186
The patch adds a possibility to make library calls on NVPTX.
An important thing about library functions - they must be defined within
the current module. This basically should guarantee that we produce a
valid PTX assembly (without calls to not defined functions). The one who
wants to use the libcalls is probably will have to link against
compiler-rt or any other implementation.
Currently, it's completely impossible to make library calls because of
error LLVM ERROR: Cannot select: i32 = ExternalSymbol '...'. But we can
lower ExternalSymbol to TargetExternalSymbol and verify if the function
definition is available.
Also, there was an issue with a DAG during legalisation. When we expand
instruction into libcall, the inner call-chain isn't being "integrated"
into outer chain. Since the last "data-flow" (call retval load) node is
located in call-chain earlier than CALLSEQ_END node, the latter becomes
a leaf and therefore a dead node (and is being removed quite fast).
Proposed here solution relies on another data-flow pseudo nodes
(ProxyReg) which purpose is only to keep CALLSEQ_END at legalisation and
instruction selection phases - we remove the pseudo instructions before
register scheduling phase.
Patch by Denys Zariaiev!
Differential Revision: https://reviews.llvm.org/D34708
llvm-svn: 350069
This is an alternative to what I attempted in D56057.
GetDemandedBits is a special version of SimplifyDemandedBits that allows simplifications even when the operand has other uses. GetDemandedBits will only do simplifications that allow a node to be bypassed. It won't create new nodes or alter any of the other users.
I had to add support for bypassing SIGN_EXTEND_INREG to GetDemandedBits.
Based on a patch that Simon Pilgrim sent me in email.
Fixes PR40142.
llvm-svn: 350059
More migration so we can disable the implicit int -> LocationSize
conversion.
All of these are either scatter/gather'ed vector instructions, or direct
loads. Hence, they're all precise.
Perhaps if we see way more getTypeStoreSize calls, we can make a
getTypeStoreLocationSize (or similar) as a wrapper that applies this
::precise. Doesn't appear that it's a good idea to make getTypeStoreSize
return a LocationSize itself, however.
llvm-svn: 350042
It's dangerous to knowingly create an illegal vector type
no matter what stage of combining we're in.
This prevents the missed folding/scalarization seen in:
https://bugs.llvm.org/show_bug.cgi?id=40146
llvm-svn: 350034
trunc (add X, C ) --> add (trunc X), C'
If we're throwing away the top bits of an 'add' instruction, do it in the narrow destination type.
This makes the truncate-able opcode list identical to the sibling transform done in IR (in instcombine).
This change used to show regressions for x86, but those are gone after D55494.
This gets us closer to deleting the x86 custom function (combineTruncatedArithmetic)
that does almost the same thing.
Differential Revision: https://reviews.llvm.org/D55866
llvm-svn: 350006
This saves materializing the immediate. The additional forms are less
common (they don't usually show up for bitfield insert/extract), but
they're still relevant.
I had to add a new target hook to prevent DAGCombine from reversing the
transform. That isn't the only possible way to solve the conflict, but
it seems straightforward enough.
Differential Revision: https://reviews.llvm.org/D55630
llvm-svn: 349857
This patch enables funnel shift -> rotate building for all ROTL/ROTR custom/legal operations.
AFAICT X86 was the last target that was missing modulo support (PR38243), but I've tried to CC stakeholders for every target that has ROTL/ROTR custom handling for their final OK.
Differential Revision: https://reviews.llvm.org/D55747
llvm-svn: 349765
Now that SimplifyDemandedBits/SimplifyDemandedVectorElts is simplifying vector elements, we're seeing more constant BUILD_VECTOR containing undefs.
This patch provides opt-in support for UNDEF elements in matchBinaryPredicate, passing NULL instead of the result ConstantSDNode* argument.
I've updated the (or (and X, c1), c2) -> (and (or X, c2), c1|c2) fold to demonstrate its use, which I believe is safe for undef cases.
Differential Revision: https://reviews.llvm.org/D55822
llvm-svn: 349629
Now that SimplifyDemandedBits/SimplifyDemandedVectorElts is simplifying vector elements, we're seeing more constant BUILD_VECTOR containing undefs.
This patch provides opt-in support for UNDEF elements in matchBinaryPredicate, passing NULL instead of the result ConstantSDNode* argument.
Differential Revision: https://reviews.llvm.org/D55822
llvm-svn: 349628
As described on PR40091, we have several places where zext (and zext_vector_inreg) fold an undef input into an undef output. For zero extensions this is incorrect as the output should guarantee to least have the new upper bits set to zero.
SimplifyDemandedVectorElts is the worst offender (and its the most likely to cause new undefs to appear) but DAGCombiner's tryToFoldExtendOfConstant has a similar issue.
Thanks to @dmgreen for catching this.
Differential Revision: https://reviews.llvm.org/D55883
llvm-svn: 349625
Now that SimplifyDemandedBits/SimplifyDemandedVectorElts are simplifying vector elements, we're seeing more constant BUILD_VECTOR containing UNDEFs.
This patch provides opt-in handling of UNDEF elements in matchUnaryPredicate, passing NULL instead of the ConstantSDNode* argument.
I've updated SelectionDAG::simplifyShift to demonstrate its use.
Differential Revision: https://reviews.llvm.org/D55819
llvm-svn: 349616
SelectionDAG currently changes these intrinsics to function calls, but that won't work
for other ISel's. Also we want to eventually support nonlazybind and weak linkage coming
from the front-end which we can't do in SelectionDAG.
llvm-svn: 349552
For opcodes not covered by SimplifyDemandedVectorElts, SimplifyDemandedBits might be able to help now that it supports demanded elts as well.
llvm-svn: 349466
The assertion type is always supposed to be a scalar type. So if the result VT of the assertion is a vector, we need to get the scalar VT before we can compare them.
Similarly for the assert above it.
I don't have a test case because I don't know of any place we violate this today. A coworker found this while trying to use r347287 on the 6.0 branch without also having r336868
llvm-svn: 349390
This is an initial patch to add the necessary support for a DemandedElts argument to SimplifyDemandedBits, more closely matching computeKnownBits and to help improve vector codegen.
I've added only a small amount of the changes necessary to get at least one test to update - a lot more can be done but I'd like to add these methodically with proper test coverage, at the same time the hope is to slowly move some/all of SimplifyDemandedVectorElts into SimplifyDemandedBits as well.
Differential Revision: https://reviews.llvm.org/D55768
llvm-svn: 349374
We keep a few iterators into the basic block we're selecting while
performing FastISel. Usually this is fine, but occasionally code wants
to remove already-emitted instructions. When this happens we have to be
careful to update those iterators so they're not pointint at dangling
memory.
llvm-svn: 349365
The transform performs a bitwise logic op in a wider type followed by
truncate when both inputs are truncated from the same source type:
logic_op (truncate x), (truncate y) --> truncate (logic_op x, y)
There are a bunch of other checks that should prevent doing this when
it might be harmful.
We already do this transform for scalars in this spot. The vector
limitation was shared with a check for the case when the operands are
extended. I'm not sure if that limit is needed either, but that would
be a separate patch.
Differential Revision: https://reviews.llvm.org/D55448
llvm-svn: 349303
Also exposes an issue in DAGCombiner::visitFunnelShift where we were assuming the shift amount had the result type (after legalization it'll have the targets shift amount type).
llvm-svn: 349298
Summary:
If the setcc already has the target desired type we can reach the getSetCC/getSExtOrTrunc after the MatchingVecType check with the exact same types as the nodes we started with. This causes those causes VsetCC to be CSEd to N0 and the getSExtOrTrunc will CSE to N. When we return N, the caller will think that meant we called CombineTo and did our own worklist management. But that's not what happened. This prevents target hooks from being called for the node.
To fix this, I've now returned SDValue if the setcc is already the desired type. But to avoid some regressions in X86 I've had to disable one of the target combines that wasn't being reached before in the case of a (sext (setcc)). If we get vector widening legalization enabled that entire function will be deleted anyway so hopefully this is only for the short term.
Reviewers: RKSimon, spatel
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D55459
llvm-svn: 349137
This isn't quite NFC, but I don't know how to expose
any outward diffs from these changes. Mostly, this
was confusing because it used 'VT' to refer to the
operand type rather the usual type of the input node.
There's also a large block at the end that is dedicated
solely to matching loads, but that wasn't obvious. This
could probably be split up into separate functions to
make it easier to see.
It's still not clear to me when we make certain transforms
because the legality and constant conditions are
intertwined in a way that might be improved.
llvm-svn: 349095
This is a retry of rL349051 (reverted at rL349056). I changed the check for dead-ness from
number of uses to an opcode test for DELETED_NODE based on existing similar code.
Differential Revision: https://reviews.llvm.org/D55655
llvm-svn: 349058
Move existing rotation expansion code into TargetLowering and set it up for vectors as well.
Ideally this would share more of the funnel shift expansion, but we handle the shift amount modulo quite differently at the moment.
Begun removing x86 vector rotate custom lowering to use the expansion.
llvm-svn: 349025
Summary:
All targets either just return false here or properly model `Fast`, so I
don't think there is any reason to prevent CodeGen from doing the right
thing here.
Subscribers: nemanjai, javed.absar, eraman, jsji, llvm-commits
Differential Revision: https://reviews.llvm.org/D55365
llvm-svn: 349016
This patch introduces a generic function to determine whether a given vector type is known to be a splat value for the specified demanded elements, recursing up the DAG looking for BUILD_VECTOR or VECTOR_SHUFFLE splat patterns.
It also keeps track of the elements that are known to be UNDEF - it returns true if all the demanded elements are UNDEF (as this may be useful under some circumstances), so this needs to be handled by the caller.
A wrapper variant is also provided that doesn't take the DemandedElts or UndefElts arguments for cases where we just want to know if the SDValue is a splat or not (with/without UNDEFS).
I had hoped to completely remove the X86 local version of this function, but I'm seeing some regressions in shift/rotate codegen that will take a little longer to fix and I hope to get this in sooner so I can continue work on PR38243 which needs more capable splat detection.
Differential Revision: https://reviews.llvm.org/D55426
llvm-svn: 348953
If either of the operand elements are zero then we know the result element is going to be zero (even if the other element is undef).
Differential Revision: https://reviews.llvm.org/D55558
llvm-svn: 348926
Add an intrinsic that takes 2 signed integers with the scale of them provided
as the third argument and performs fixed point multiplication on them.
This is a part of implementing fixed point arithmetic in clang where some of
the more complex operations will be implemented as intrinsics.
Differential Revision: https://reviews.llvm.org/D54719
llvm-svn: 348912
Summary:
All targets either just return false here or properly model `Fast`, so I
don't think there is any reason to prevent CodeGen from doing the right
thing here.
Subscribers: nemanjai, javed.absar, eraman, jsji, llvm-commits
Differential Revision: https://reviews.llvm.org/D55365
llvm-svn: 348843
If all the demanded elements of the SimplifyDemandedVectorElts are known to be UNDEF, we can simplify to an ISD::UNDEF node.
Zero constant folding will be handled in a future patch - its a little trickier as we often have bitcasted zero values.
Differential Revision: https://reviews.llvm.org/D55511
llvm-svn: 348784
As discussed on D55511, this caused an issue if the inner node deletes a node that the outer node depends upon. As it doesn't affect any lit-tests and I've only been able to expose this with the D55511 change I'm committing this now.
llvm-svn: 348781
This triggers an assert when combining concat_vectors of a bitcast of
merge_values.
With asserts disabled, it fails to select:
fatal error: error in backend: Cannot select: 0x7ff19d000e90: i32 = any_extend 0x7ff19d000ae8
0x7ff19d000ae8: f64,ch = CopyFromReg 0x7ff19d000c20:1, Register:f64 %1
0x7ff19d000b50: f64 = Register %1
In function: d
Differential Revision: https://reviews.llvm.org/D55507
llvm-svn: 348759
Currently, dbg.value's of "nullptr" are dropped when entering a SelectionDAG --
apparently just because of an oversight when recognising Values that are
constant (see PR39787). This patch adds ConstantPointerNull to the list of
constants that can be turned into DBG_VALUEs.
The matter of what bit-value a null pointer constant in LLVM has was raised
in this mailing list thread:
http://lists.llvm.org/pipermail/llvm-dev/2018-December/128234.html
Where it transpires LLVM relies on (IR) null pointers being zero valued,
thus I've baked this assumption into the patch.
Differential Revision: https://reviews.llvm.org/D55227
llvm-svn: 348753
This is a fix for PR39896, where dbg.value's of SDNodes that have been
optimised out do not lead to "DBG_VALUE undef" instructions being created.
Such undef instructions are necessary to terminate earlier variable
ranges, otherwise variable values leak past the point where they're valid.
The "invalidated" flag of SDDbgValue is currently being abused to mean two
things:
* The corresponding SDNode is now invalid
* This SDDbgValue should not be emitted
Of which there are several legitimate combinations of meaning:
* The SDNode has been invalidated and we should emit "DBG_VALUE undef"
* The SDNode has been invalidated but the debug data was salvaged, don't
emit anything for this SDDbgValue
* This SDDbgValue has been emitted
This patch introduces distinct "Emitted" and "Invalidated" fields to the
SDDbgValue class, updates users accordingly, and generates "undef"
DBG_VALUEs for invalidated records. Awkwardly, there are circumstances
where we emit SDDbgValue's twice, specifically DebugInfo/X86/dbg-addr-dse.ll
which I've preserved.
Differential Revision: https://reviews.llvm.org/D55372
llvm-svn: 348751
This is effectively re-committing the changes from:
rL347917 (D54640)
rL348195 (D55126)
...which were effectively reverted here:
rL348604
...because the code had a bug that could induce infinite looping
or eventual out-of-memory compilation.
The bug was that this code did not guard against transforming
opaque constants. More details are in the post-commit mailing
list thread for r347917. A reduced test for that is included
in the x86 bool-math.ll file. (I wasn't able to reduce a PPC
backend test for this, but it was almost the same pattern.)
Original commit message for r347917:
The motivating case for this is shown in:
https://bugs.llvm.org/show_bug.cgi?id=32023
and the corresponding rot16.ll regression tests.
Because x86 scalar shift amounts are i8 values, we can end up with trunc-binop-trunc
sequences that don't get folded in IR.
As the TODO comments suggest, there will be regressions if we extend this (for x86,
we mostly seem to be missing LEA opportunities, but there are likely vector folds
missing too). I think those should be considered existing bugs because this is the
same transform that we do as an IR canonicalization in instcombine. We just need
more tests to make those visible independent of this patch.
llvm-svn: 348706
These nodes should have two results. A real VT and a Glue. But this code would have returned Undef which would only be a single result. But we're in the single result version of getNode so these opcodes should never be seen by this function anyway.
llvm-svn: 348670
As discussed in the post-commit thread of r347917, this
transform is fighting with an existing transform causing
an infinite loop or out-of-memory, so this is effectively
reverting r347917 and its follow-up r348195 while we
investigate the bug.
llvm-svn: 348604
If this is not a valid way to assign an SDLoc, then we get this
wrong all over SDAG.
I don't know enough about the SDAG to explain this. IIUC, theoretically,
debug info is not supposed to affect codegen. But here it has clearly
affected 3 different targets, and the x86 change is an actual improvement.
llvm-svn: 348552
We shouldn't care about the debug location for a node that
we're creating, but attaching the root of the pattern should
be the best effort. (If this is not true, then we are doing
it wrong all over the SDAG).
This is no-functional-change-intended, and there are no
regression test diffs...and that's what I expected. But
there's a similar line above this diff, where those
assumptions apparently do not hold.
llvm-svn: 348550
This was probably organized as it was because bswap is a unary op.
But that's where the similarity to the other opcodes ends. We should
not limit this transform to scalars, and we should not try it if
either input has other uses. This is another step towards trying to
clean this whole function up to prevent it from causing infinite loops
and memory explosions.
Earlier commits in this series:
rL348501
rL348508
rL348518
llvm-svn: 348534
Unlike some of the folds in hoistLogicOpWithSameOpcodeHands()
above this shuffle transform, this has the expected hasOneUse()
checks in place.
llvm-svn: 348523
This patch introduces a new DAGCombiner rule to simplify concat_vectors nodes:
concat_vectors( bitcast (scalar_to_vector %A), UNDEF)
--> bitcast (scalar_to_vector %A)
This patch only partially addresses PR39257. In particular, it is enough to fix
one of the two problematic cases mentioned in PR39257. However, it is not enough
to fix the original test case posted by Craig; that particular case would
probably require a more complicated approach (and knowledge about used bits).
Before this patch, we used to generate the following code for function PR39257
(-mtriple=x86_64 , -mattr=+avx):
vmovsd (%rdi), %xmm0 # xmm0 = mem[0],zero
vxorps %xmm1, %xmm1, %xmm1
vblendps $3, %xmm0, %xmm1, %xmm0 # xmm0 = xmm0[0,1],xmm1[2,3]
vmovaps %ymm0, (%rsi)
vzeroupper
retq
Now we generate this:
vmovsd (%rdi), %xmm0 # xmm0 = mem[0],zero
vmovaps %ymm0, (%rsi)
vzeroupper
retq
As a side note: that VZEROUPPER is completely redundant...
I guess the vzeroupper insertion pass doesn't realize that the definition of
%xmm0 from vmovsd is already zeroing the upper half of %ymm0. Note that on
%-mcpu=btver2, we don't get that vzeroupper because pass vzeroupper insertion
%pass is disabled.
Differential Revision: https://reviews.llvm.org/D55274
llvm-svn: 348522
The PPC test with 2 extra uses seems clearly better by avoiding this transform.
With 1 extra use, we also prevent an extra register move (although that might
be an RA problem). The general rule should be to only make a change here if
it is always profitable. The x86 diffs are all neutral.
llvm-svn: 348518
The AVX512 diffs are neutral, but the bswap test shows a clear overreach in
hoistLogicOpWithSameOpcodeHands(). If we don't check for other uses, we can
increase the instruction count.
This could also fight with transforms trying to go in the opposite direction
and possibly blow up/infinite loop. This might be enough to solve the bug
noted here:
http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20181203/608593.html
I did not add the hasOneUse() checks to all opcodes because I see a perf
regression for at least one opcode. We may decide that's irrelevant in the
face of potential compiler crashing, but I'll see if I can salvage that first.
llvm-svn: 348508
Because we're potentially peeking through a bitcast in this transform,
we need to use overall bitwidths rather than number of elements to
determine when it's safe to proceed.
Should fix:
https://bugs.llvm.org/show_bug.cgi?id=39893
llvm-svn: 348383
This is an initial patch to add a minimum level of support for funnel shifts to the SelectionDAG and to begin wiring it up to the X86 SHLD/SHRD instructions.
Some partial legalization code has been added to handle the case for 'SlowSHLD' where we want to expand instead and I've added a few DAG combines so we don't get regressions from the existing DAG builder expansion code.
Differential Revision: https://reviews.llvm.org/D54698
llvm-svn: 348353
Fix potential issue with the ISD::INSERT_VECTOR_ELT case tweaking the DemandedElts mask instead of using a local copy - so later uses of the mask use the tweaked version.....
Noticed while investigating adding zero/undef folding to SimplifyDemandedVectorElts and the altered DemandedElts mask was causing mismatches.
llvm-svn: 348348
There's a 64k limit on the number of SDNode operands, and some very large
functions with 64k or more loads can cause crashes due to this limit being hit
when a TokenFactor with this many operands is created. To fix this, create
sub-tokenfactors if we've exceeded the limit.
No test case as it requires a very large function.
rdar://45196621
Differential Revision: https://reviews.llvm.org/D55073
llvm-svn: 348324
PR17686 demonstrates that for some targets FP exceptions can fire in cases where the FP_TO_UINT is expanded using a FP_TO_SINT instruction.
The existing code converts both the inrange and outofrange cases using FP_TO_SINT and then selects the result, this patch changes this for 'strict' cases to pre-select the FP_TO_SINT input and the offset adjustment.
The X87 cases don't need the strict flag but generates much nicer code with it....
Differential Revision: https://reviews.llvm.org/D53794
llvm-svn: 348251
Add support for ISD::*_EXTEND and ISD::*_EXTEND_VECTOR_INREG opcodes.
The extra broadcast in trunc-subvector.ll will be fixed in an upcoming patch.
llvm-svn: 348246
This is the smallest vector enhancement I could find to D54640.
Here, we're allowing narrowing to only legal vector ops because we'll see
regressions without that. All of the test diffs are wins from what I can tell.
With AVX/AVX512, we can shrink ymm/zmm ops to xmm.
x86 vector multiplies are the problem case that we're avoiding due to the
patchwork ISA, and it's not clear to me if we can dance around those
regressions using TLI hooks or if we need preliminary patches to plug those
holes.
Differential Revision: https://reviews.llvm.org/D55126
llvm-svn: 348195
Summary:
Under -x86-experimental-vector-widening-legalization, fp_to_uint/fp_to_sint with a smaller than 128 bit vector type results are custom type legalized by promoting the result to a 128 bit vector by promoting the elements, inserting an assertzext/assertsext, then truncating back to original type. The truncate will be further legalizdd to a pack shuffle. In the case of a v8i8 result type, we'll end up with a v8i16 fp_to_sint. This will need to be further legalized during vector op legalization by promoting to v8i32 and then truncating again. Under avx2 this produces good code with two pack instructions, but Under avx512 this will result in a truncate instruction and a packuswb instruction. But we should be able to get away with a single truncate instruction.
The other option is to promote all the way to vXi32 result type during the first type legalization. But in some experimentation that seemed to require more work to produce good code for other configurations.
Reviewers: RKSimon, spatel
Reviewed By: RKSimon
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D54836
llvm-svn: 348158
This makes the SDAG behavior consistent with the way we do this in IR.
It's possible that we were getting the wrong answer before. For example,
'xor undef, undef --> 0' but 'xor undef, C' --> undef.
But the most practical improvement is likely as shown in the tests here -
for FP, we were overconstraining undef lanes to NaN, and that can prevent
vector simplifications/narrowing (see D51553).
llvm-svn: 348090
This change prevents the crash noted in the post-commit comments
for rL347478 :
http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20181119/605166.html
We can't guarantee that an oversized shift amount is folded away,
so we have to check for it.
Note that I committed an incomplete fix for that crash with:
rL347502
But as discussed here:
http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20181126/605679.html
...we have to try harder.
So I'm not sure how to expose the bug now (and apparently no fuzzers have found
a way yet either).
On the plus side, we have discovered that we're missing real optimizations by
not simplifying nodes sooner, so the earlier fix still has value, and there's
likely more value in extending that so we can simplify more opcodes and simplify
when doing RAUW and/or putting nodes on the combiner worklist.
Differential Revision: https://reviews.llvm.org/D54954
llvm-svn: 348089
D52935 introduced the ability for SimplifyDemandedBits to call SimplifyDemandedVectorElts through BITCASTs if the demanded bit mask entirely covered the sub element.
This patch relaxes this to demanding an element if we need any bit from it.
Differential Revision: https://reviews.llvm.org/D54761
llvm-svn: 348073
Summary:
The VirtReg2Value mapping is crucial for getting consistently
reliable divergence information into the SelectionDAG. This
patch fixes a bunch of issues that lead to incorrect divergence
info and introduces tight assertions to ensure we don't regress:
1. VirtReg2Value is generated lazily; there were some cases where
a lookup was performed before all relevant virtual registers were
created, leading to an out-of-sync mapping. Those cases were:
- Complex code to lower formal arguments that generated CopyFromReg
nodes from live-in registers (fixed by never querying the mapping
for live-in registers).
- Code that generates CopyToReg for formal arguments that are used
outside the entry basic block (fixed by never querying the
mapping for Register nodes, which don't need the divergence info
anyway).
2. For complex values that are lowered to a sequence of registers,
all registers must be reflected in the VirtReg2Value mapping.
I am not adding any new tests, since I'm not actually aware of any
bugs that these problems are causing with trunk as-is. However,
I recently added a test case (in r346423) which fails when D53283 is
applied without this change. Also, the new assertions should provide
most of the effective test coverage.
There is one test change in sdwa-peephole.ll. The underlying issue
is that since the divergence info is now correct, the DAGISel will
select V_OR_B32 directly instead of S_OR_B32. This leads to an extra
COPY which affects the behavior of MachineLICM in a way that ends up
with the S_MOV_B32 with the constant in a different basic block than
the V_OR_B32, which is presumably what defeats the peephole.
Reviewers: alex-t, arsenm, rampitec
Subscribers: kzhuravl, jvesely, wdng, yaxunl, dstuttard, tpr, t-tye, llvm-commits
Differential Revision: https://reviews.llvm.org/D54340
llvm-svn: 348049
Summary:
If a given liveness arg of STATEPOINT is at a fixed frame index
(e.g. a function argument passed on stack), prefer to use this
fixed location even the address is also in a register. If we use
the register it will generate a spill, which is not necessary
since the fixed frame index can be directly recorded in the stack
map.
Patch by Cherry Zhang <cherryyz@google.com>.
Reviewers: thanm, niravd, reames
Reviewed By: reames
Subscribers: cherryyz, reames, anna, arphaman, llvm-commits
Differential Revision: https://reviews.llvm.org/D53889
llvm-svn: 347998
Summary:
This simplifies writing predicates for pattern fragments that are
automatically re-associated or commuted.
For example, a followup patch adds patterns for fragments of the form
(add (shl $x, $y), $z) to the AMDGPU backend. Such patterns are
automatically commuted to (add $z, (shl $x, $y)), which makes it basically
impossible to refer to $x, $y, and $z generically in the PredicateCode.
With this change, the PredicateCode can refer to $x, $y, and $z simply
as `Operands[i]`.
Test confirmed that there are no changes to any of the generated files
when building all (non-experimental) targets.
Change-Id: I61c00ace7eed42c1d4edc4c5351174b56b77a79c
Reviewers: arsenm, rampitec, RKSimon, craig.topper, hfinkel, uweigand
Subscribers: wdng, tpr, llvm-commits
Differential Revision: https://reviews.llvm.org/D51994
llvm-svn: 347992
For targets where i32 is not a legal type (e.g. 64-bit RISC-V),
LegalizeIntegerTypes must promote the result of ISD::FLT_ROUNDS_.
Differential Revision: https://reviews.llvm.org/D53820
llvm-svn: 347986
For targets where i32 is not a legal type (e.g. 64-bit RISC-V),
LegalizeIntegerTypes must promote the operands of ISD::PREFETCH.
Differential Revision: https://reviews.llvm.org/D53281
llvm-svn: 347980
For targets where i32 is not a legal type (e.g. 64-bit RISC-V),
LegalizeIntegerTypes must promote the operand.
Differential Revision: https://reviews.llvm.org/D53279
llvm-svn: 347978
DAGTypeLegalizer::PromoteSetCCOperands currently prefers to zero-extend
operands when it is able to do so. For some targets this is more expensive
than a sign-extension, which is also a valid choice. Introduce the
isSExtCheaperThanZExt hook and use it in the new SExtOrZExtPromotedInteger
helper. On RISC-V, we prefer sign-extension for FromTy == MVT::i32 and ToTy ==
MVT::i64, as it can be performed using a single instruction.
Differential Revision: https://reviews.llvm.org/D52978
llvm-svn: 347977
The motivating case for this is shown in:
https://bugs.llvm.org/show_bug.cgi?id=32023
and the corresponding rot16.ll regression tests.
Because x86 scalar shift amounts are i8 values, we can end up with trunc-binop-trunc
sequences that don't get folded in IR.
As the TODO comments suggest, there will be regressions if we extend this (for x86,
we mostly seem to be missing LEA opportunities, but there are likely vector folds
missing too). I think those should be considered existing bugs because this is the
same transform that we do as an IR canonicalization in instcombine. We just need
more tests to make those visible independent of this patch.
Differential Revision: https://reviews.llvm.org/D54640
llvm-svn: 347917
I believe we should be legalizing these with the rest of vector binary operations. If any custom lowering is required for these nodes, this will give the DAG combine between LegalizeVectorOps and LegalizeDAG to run on the custom code before constant build_vectors are lowered in LegalizeDAG.
I've moved MULHU/MULHS handling in AArch64 from Lowering to isel. Moving the lowering earlier caused build_vector+extract_subvector simplifications to kick in which made the generated code worse.
Differential Revision: https://reviews.llvm.org/D54276
llvm-svn: 347902
SplitVecOp_TruncateHelper tries to promote the result type while splitting FP_TO_SINT/UINT. It then concatenates the result and introduces a truncate to the original result type. But it does this without inserting the AssertZExt/AssertSExt that the regular result type promotion would insert. Nor does it turn FP_TO_UINT into FP_TO_SINT the way normal result type promotion for these operations does. This is bad on X86 which doesn't support FP_TO_SINT until AVX512.
This patch disables the use of SplitVecOp_TruncateHelper for these operations and just lets normal promotion handle it. I've tweaked a couple things in X86ISelLowering to avoid a few obvious regressions there. I believe all the changes on X86 are improvements. The other targets look neutral.
Differential Revision: https://reviews.llvm.org/D54906
llvm-svn: 347593
We might find a target specific node that needs to be unwrapped after we look through an add/or. Otherwise we get inconsistent results if one pointer is just X86WrapperRIP and the other is (add X86WrapperRIP, C)
Differential Revision: https://reviews.llvm.org/D54818
llvm-svn: 347591
This should likely be adjusted to limit this transform
further, but these diffs should be clear wins.
If we have blendv/conditional move, then we should assume
those are cheap ops. The loads become independent of the
compare, so those can be speculated before we need to use
the values in the blend/mov.
llvm-svn: 347526
rL347502 moved the null sibling, so we should group all of these
together. I'm not sure why these aren't methods of the SDValue
class itself, but that's another patch if that's possible.
llvm-svn: 347523
...and use them to avoid creating obviously undef values as
discussed in the post-commit thread for r347478.
The diffs in vector div/rem show that we were missing real
optimizations by creating bogus shift nodes.
llvm-svn: 347502
This code takes a truncate, fp_to_int, or int_to_fp with a legal result type and an input type that needs to be split and enlarges the elements in the result type before doing the split. Then inserts a follow up truncate or fp_round after concatenating the two halves back together.
But if the input type of the original op is being split on its way to ultimately being scalarized we're just going to end up building a vector from scalars and then truncating or rounding it in the vector register. Seems kind of silly to enlarge the result element type of the operation only to end up with scalar code and then building a vector with large elements only to make the elements smaller again in the vector register. Seems better to just try to get away producing smaller result types in the scalarized code.
The X86 test case that changes is a pretty contrived test case that exists because of a bug we used to have in our AVG matching code. I think the code is better now, but its not realistic anyway.
llvm-svn: 347482
SplitVecOp_TruncateHelper tries to introduce a multilevel truncate to avoid scalarization. But if splitting the result type would still be a legal type we don't need to do that.
The comment block at the top of the function implied that this was already implemented. I looked back through the history and it doesn't look to have ever been checked.
llvm-svn: 347479
We fail to canonicalize IR this way (prefer 'not' ops to arbitrary 'xor'),
but that would not matter without this patch because DAGCombiner was
reversing that transform. I think we need this transform in the backend
regardless of what happens in IR to catch cases where the shift-xor
is formed late from GEP or other ops.
https://rise4fun.com/Alive/NC1
Name: shl
Pre: (-1 << C2) == C1
%shl = shl i8 %x, C2
%r = xor i8 %shl, C1
=>
%not = xor i8 %x, -1
%r = shl i8 %not, C2
Name: shr
Pre: (-1 u>> C2) == C1
%sh = lshr i8 %x, C2
%r = xor i8 %sh, C1
=>
%not = xor i8 %x, -1
%r = lshr i8 %not, C2
https://bugs.llvm.org/show_bug.cgi?id=39657
llvm-svn: 347478
This transform needs to be limited.
We are converting to a constant pool load very early, and we
are turning loads that are independent of the select condition
(and therefore speculatable) into a dependent non-speculatable
load.
We may also be transferring a condition code from an FP register
to integer to create that dependent load.
llvm-svn: 347424
This is another step in vector narrowing - a follow-up to D53784
(and hoping to eventually squash potential regressions seen in
D51553).
The x86 test diffs are wins, but the AArch64 diff is probably not.
That problem already exists independent of this patch (see PR39722), but it
went unnoticed in the previous patch because there were no regression tests
that showed the possibility.
The x86 diff in i64-mem-copy.ll is close. Given the frequency throttling
concerns with using wider vector ops, an extra extract to reduce vector
width is the right trade-off at this level of codegen.
Differential Revision: https://reviews.llvm.org/D54392
llvm-svn: 347356
This uncovered an off-by-one typo in SimplifyDemandedVectorElts's INSERT_SUBVECTOR handling as its bounds check was bailing on safe indices.
llvm-svn: 347313
For bitcast nodes from larger element types, add the ability for SimplifyDemandedVectorElts to call SimplifyDemandedBits by merging the elts mask to a bits mask.
I've raised https://bugs.llvm.org/show_bug.cgi?id=39689 to deal with the few places where SimplifyDemandedBits's lack of vector handling is a problem.
Differential Revision: https://reviews.llvm.org/D54679
llvm-svn: 347301
Summary:
We already support this for scalars, but it was explicitly disabled for vectors. In the updated test cases this allows us to see the upper bits are zero to use less multiply instructions to emulate a 64 bit multiply.
This should help with this ispc issue that a coworker pointed me to https://github.com/ispc/ispc/issues/1362
Reviewers: spatel, efriedma, RKSimon, arsenm
Reviewed By: spatel
Subscribers: wdng, llvm-commits
Differential Revision: https://reviews.llvm.org/D54725
llvm-svn: 347287
Consistently use (!LegalOperations || isOperationLegalOrCustom) for all node pairs.
Differential Revision: https://reviews.llvm.org/D53478
llvm-svn: 347255
As discussed on D53794, for float types with ranges smaller than the destination integer type, then we should be able to just use a regular FP_TO_SINT opcode.
I thought we'd need to provide MSA test cases for very small integer types as well (fp16 -> i8 etc.), but it turns out that promotion will kick in so they're unnecessary.
Differential Revision: https://reviews.llvm.org/D54703
llvm-svn: 347251
Sadly, this duplicates (twice) the logic from InstSimplify. There
might be some way to at least share the DAG versions of the code,
but copying the folds seems to be the standard method to ensure
that we don't miss these folds.
Unlike in IR, we don't run DAGCombiner to fixpoint, so there's no
way to ensure that we do these kinds of simplifications unless the
code is repeated at node creation time and during combines.
There were other tests that would become worthless with this
improvement that I changed as pre-commits:
rL347161
rL347164
rL347165
rL347166
rL347167
I'm not sure how to salvage the remaining tests (diffs in this patch).
So the x86 tests verify that the new code is working as intended.
The AMDGPU test is actually similar to my motivating case: we have
some undef value that has survived to machine IR in an x86 test, and
then it gets folded in some weird way, or we crash if we don't transfer
the undef flag. But we would have been better off never getting to that
point by doing these simplifications.
This will lead back to PR32023 someday...
https://bugs.llvm.org/show_bug.cgi?id=32023
llvm-svn: 347170
For example, on X86 we emit a sign_extend_vector_inreg from LowerLoad and without sse4.1 this node will need further legalization. Previously this sign_extend_vector_inreg was being custom lowered during DAG legalization instead of vector op legalization.
Unfortunately, this doesn't seem to matter for the output of any existing lit tests.
llvm-svn: 347094
Legalizer used to request an ext load from i8 to i1 when promoting
vector element type to i8. Fixed.
Differential Revision: https://reviews.llvm.org/D54440
llvm-svn: 346795
Previously, the extend_vector_inreg opcode required their input register to be the same total width as their output. But this doesn't match up with how the X86 instructions are defined. For X86 the input just needs to be a legal type with at least enough elements to cover the output.
This patch weakens the check on these nodes and allows them to be used as long as they have more input elements than output elements. I haven't changed type legalization behavior so it will still create them with matching input and output sizes.
X86 will custom legalize these nodes by shrinking the input to be a 128 bit vector and once we've done that we treat them as legal operations. We still have one case during type legalization where we must custom handle v64i8 on avx512f targets without avx512bw where v64i8 isn't a legal type. In this case we will custom type legalize to a *extend_vector_inreg with a v16i8 input. After that the input is a legal type so type legalization should ignore the node and doesn't need to know about the relaxed restriction. We are no longer allowed to use the default expansion for these nodes during vector op legalization since the default expansion uses a shuffle which required the widths to match. Custom legalization for all types will prevent us from reaching the default expansion code.
I believe DAG combine works correctly with the released restriction because it doesn't check the number of input elements.
The rest of the patch is changing X86 to use either the vector_inreg nodes or the regular zero_extend/sign_extend nodes. I had to add additional isel patterns to handle any_extend during isel since simplifydemandedbits can create them at any time so we can't legalize to zero_extend before isel. We don't yet create any_extend_vector_inreg in simplifydemandedbits.
Differential Revision: https://reviews.llvm.org/D54346
llvm-svn: 346784
The IEEE-754 Standard makes it clear that fneg(x) and
fsub(-0.0, x) are two different operations. The former is a bitwise
operation, while the latter is an arithmetic operation. This patch
creates a dedicated FNeg IR Instruction to model that behavior.
Differential Revision: https://reviews.llvm.org/D53877
llvm-svn: 346774
It should be ok to create a new build_vector after legal operations so long as it doesn't cause an infinite loop in DAG combiner.
Unfortunately, X86's custom constant folding in combineVSZext is hiding any test changes from this. But I'm trying to get to a point where that X86 specific code isn't necessary at all.
Differential Revision: https://reviews.llvm.org/D54285
llvm-svn: 346728
Summary:
Handle extra output from index loads in cases where we wish to
forward a load value directly from a preceeding store.
Fixes PR39571.
Reviewers: peter.smith, rengolin
Subscribers: javed.absar, hiraditya, arphaman, llvm-commits
Differential Revision: https://reviews.llvm.org/D54265
llvm-svn: 346654
This is a long-awaited follow-up suggested in D33578. Since then, we've picked up even more
opportunities for vector narrowing from changes like D53784, so there are a lot of test diffs.
Apart from 2-3 strange cases, these are all wins.
I've structured this to be no-functional-change-intended for any target except for x86
because I couldn't tell if AArch64, ARM, and AMDGPU would improve or not. All of those
targets have existing regression tests (4, 4, 10 files respectively) that would be
affected. Also, Hexagon overrides the shouldReduceLoadWidth() hook, but doesn't show
any regression test diffs. The trade-off is deciding if an extra vector load is better
than a single wide load + extract_subvector.
For x86, this is almost always better (on paper at least) because we often can fold
loads into subsequent ops and not increase the official instruction count. There's also
some unknown -- but potentially large -- benefit from using narrower vector ops if wide
ops are implemented with multiple uops and/or frequency throttling is avoided.
Differential Revision: https://reviews.llvm.org/D54073
llvm-svn: 346595
It's possible for vector op legalization to generate a shuffle. If that happens we should give a chance for DAG combine to combine that with a build_vector input.
I also fixed a bug in combineShuffleOfScalars that was considering the number of uses on a undef input to a shuffle. We don't care how many times undef is used.
Differential Revision: https://reviews.llvm.org/D54283
llvm-svn: 346530
Previous version used type erasure through a `void* (*)()` pointer,
which triggered gcc warning and implied a lot of reinterpret_cast.
This version should make it harder to hit ourselves in the foot.
Differential revision: https://reviews.llvm.org/D54203
llvm-svn: 346522
The DAGCombiner tries to SimplifySelectCC as follows:
select_cc(x, y, 16, 0, cc) -> shl(zext(set_cc(x, y, cc)), 4)
It can't cope with the situation of reordered operands:
select_cc(x, y, 0, 16, cc)
In that case we just need to swap the operands and invert the Condition Code:
select_cc(x, y, 16, 0, ~cc)
Differential Revision: https://reviews.llvm.org/D53236
llvm-svn: 346484
FindBetterNeighborChains simulateanously improves the chain
dependencies of a chain of related stores avoiding the generation of
extra token factors. For chains longer than the GatherAllAliasDepths,
stores further down in the chain will necessarily fail, a potentially
significant waste and preventing otherwise trivial parallelization.
This patch directly parallelize the chains of stores before improving
each store. This generally improves DAG-level parallelism.
Reviewers: courbet, spatel, RKSimon, bogner, efriedma, craig.topper, rnk
Subscribers: sdardis, javed.absar, hiraditya, jrtc27, atanasyan, llvm-commits
Differential Revision: https://reviews.llvm.org/D53552
llvm-svn: 346432
This adds the llvm-side support for post-inlining evaluation of the
__builtin_constant_p GCC intrinsic.
Also fixed SCCPSolver::visitCallSite to not blow up when seeing a call
to a function where canConstantFoldTo returns true, and one of the
arguments is a struct.
Updated from patch initially by Janusz Sobczak.
Differential Revision: https://reviews.llvm.org/D4276
llvm-svn: 346322
The original code avoided creating a zero vector after type legalization, but if we're after type legalization the type we have is legal. The real hazard we need to avoid is creating a build vector after op legalization. tryFoldToZero takes care of checking for this.
llvm-svn: 346119
These methods were just wrappers around getNode with additional asserts (identical and repeated 3 times). But getNode already has a switch that can be used to hold these asserts that allows them to be shared for all 3 opcodes. This also enables checking on the places that create these nodes without using the wrappers.
The rest of the patch is just changing all callers to use getNode directly.
llvm-svn: 346087
We already have custom lowering for the AVX case in LegalizeVectorOps. So its better to keep the regular extend op around as long as possible.
I had to qualify one place in DAG combine that created illegal vector extending load operations. This change by itself had no effect on any tests which is why its included here.
I've made a few cleanups to the custom lowering. The sign extend code no longer creates an identity shuffle with undef elements. The zero extend code now emits a zero_extend_vector_inreg instead of an unpckl with a zero vector.
For the high half of the custom lowering of zero_extend/any_extend, we're now using an unpckh with a zero vector or undef. Previously we used used a pshufd to move the upper 64-bits to the lower 64-bits and then used a zero_extend_vector_inreg. I think the zero vector should require less execution resources and be smaller code size.
Differential Revision: https://reviews.llvm.org/D54024
llvm-svn: 346043
reduceBuildVecConvertToConvertBuildVec vectorizes int2float in the DAGCombiner, which means that even if the LV/SLP has decided to keep scalar code using the cost models, this will override this.
While there are cases where vectorization is necessary in the DAG (mainly due to legalization artefacts), I don't think this is the case here, we should assume that the vectorizers know what they are doing.
Differential Revision: https://reviews.llvm.org/D53712
llvm-svn: 345964
Simon Pilgrim